YouTube12 Feb 2025
2h 15m

Jeff Dean & Noam Shazeer — 25 years at Google: from PageRank to AGI

Podcast cover

Dwarkesh Patel

This podcast features an interview with Jeff Dean and Noam Shazeer, two leading figures in Google's AI development, discussing their long careers at Google and the future of AI. The conversation covers their early work on foundational AI technologies, the evolution of hardware and algorithms, and the potential for future breakthroughs in areas like longer context models and continual learning. They highlight the increasing importance of inference-time compute and the potential for AI to dramatically increase human productivity, potentially leading to a rapid acceleration in AI capabilities. A key takeaway is the prediction that AI will become exponentially more powerful in the coming years, driven by both algorithmic improvements and automated exploration of new ideas. The discussion also touches upon the ethical considerations and potential risks associated with rapidly advancing AI, emphasizing the need for responsible development and deployment.

Outlines

Part 1: Introductions and Early Experiences

Part 2: Hardware, Models, and AI's Inevitability

Part 3: Google's Evolution and Long Context

Part 4: Scaling, Inference, and Debugging

Part 5: Responsible AI and Future Directions

Part 6: Model Growth and Training

Part 7: Open Source, Career Advice, and Conclusion

Sign in to continue reading, translating and more.

Open full episode in Podwise